RE: Test harness action items

Thanks, that does answer all of my questions.

With a better understanding about what is being tested, I think you are right and having second coding examples of these variations is not necessary for the test kit. I was under the impression that AT testing was a part of this testing process, but if this is limited to browser mapping support then these focus/interface distinctions wouldn't be needed.


Bryan Garaventa
Accessibility Fellow
SSB BART Group, Inc.
bryan.garaventa@ssbbartgroup.com
415.624.2709 (o)
www.SSBBartGroup.com

-----Original Message-----
From: Joanmarie Diggs [mailto:jdiggs@igalia.com] 
Sent: Thursday, April 27, 2017 2:09 PM
To: Bryan Garaventa <bryan.garaventa@ssbbartgroup.com>
Cc: Rich Schwerdtfeger <richschwer@gmail.com>; Accessible Rich Internet Applications Working Group <public-aria@w3.org>
Subject: Re: Test harness action items

Hey Bryan.

Since you state this is a quick question, I'll start with a quick
answer: The result should be the same and there is no need for two test cases for "listbox orientation unspecified."

If that's all you need to know, feel free to stop reading. <smiles> On the other hand, if you're convinced there do need to be two tests read on.

Below you provide two versions of the test, one with the tabindex=0 on the listbox itself; the other with the tabindex=0 on the option element.
Here's my first question: What difference does that make to whether or not "vertical" is exposed on a listbox element when the value of aria-orientation has not been specified by the author?

You then state:

> Both methods are valid however the element that the AT interfaces with 
> are different.

Fair enough. But we're not testing what ATs do or what the user experience is. We are testing that if an AT wants to know the orientation of the listbox, it can get it. If focus is on the listbox, the AT just needs to ask for the orientation. If focus is on the option, the AT needs to ascend the accessibility tree until it finds the ancestor that is the listbox. Having done so, the AT can then ask for the orientation.

There is nothing in the ARIA spec or the Core AAM that says that user agents must expose the listbox orientation on the option children. And I can think of no reason why a user agent would take focus or user interaction into account when exposing the orientation of a listbox. Is there one?

Where additional test cases may be called for, with respect to user interaction and authoring, are in my opinion limited to the following:

1. The assertions are related to focus and selection.

2. There's something about the authoring which might cause the element
   which is being tested to not be included in the accessibility tree.

With respect to the first, note that I said assertions; not markup. The assertion that, given a listbox without a value for aria-orientation, user agents expose vertical orientation has nothing to do with focus or selection. Sure, listboxes have focus- and selection-related aspects to them, but we're testing orientation. And orientation is neither focus nor selection, and changes to focus and selection surely will not result in changes to orientation. (Right?)

With respect to the second, that sort of thing results from getting an unexpected result during testing, scratching your head, and digging further into the problem. For a real, concrete example, see one I created, complete with explanation and bug filed:
https://www.w3.org/wiki/ARIA_1.1_Testable_Statements#cell_aria-colspan_2_on_td_html_colspan_3_with_headers_and_border.


I hope, if you read this far, I've answered all your questions. Thanks again for your help with this stuff!
--joanie


On 04/26/2017 01:40 PM, Bryan Garaventa wrote:
> Hi,
> 
> A quick question about the testing process.
> 
>  
> 
> So for the section within the test kit: “listbox orientation 
> unspecified[”, the following markup is given:
> 
>  
> 
> if given
> 
>   <div id="test" tabindex="0" role="listbox" >
> 
>     <div role="option">option 1</div>
> 
>     <div role="option">option 2</div>
> 
> </div>
> 
> then expose role: listbox and aria-orientation="vertical" for the 
> element with id="test"
> 
>  
> 
> In regard to testing later, would you expect the same result to occur 
> for the following statement?
> 
>  
> 
> if given
> 
>   <div id="test" role="listbox" >
> 
>     <div tabindex="0" role="option">option 1</div>
> 
>     <div role="option">option 2</div>
> 
> </div>
> 
> then expose role: listbox and aria-orientation="vertical" for the 
> element with id="test"
> 
>  
> 
> Both methods are valid however the element that the AT interfaces with 
> are different. Is the test result meant to be the same or do these two 
> need to be broken out?
> 
>  
> 
> Thanks,
> 
> Bryan
> 
>  
> 
>  
> 
>  
> 
> Bryan Garaventa
> 
> Accessibility Fellow
> 
> SSB BART Group, Inc.
> 
> bryan.garaventa@ssbbartgroup.com
> 
> 415.624.2709 (o)
> 
> www.SSBBartGroup.com
> 
>  
> 
> *From:* Rich Schwerdtfeger [mailto:richschwer@gmail.com]
> *Sent:* Tuesday, April 25, 2017 1:43 PM
> *To:* Bryan Garaventa <bryan.garaventa@ssbbartgroup.com>
> *Cc:* Joanmarie Diggs <jdiggs@igalia.com>; Mike Cooper 
> <cooper@w3.org>; Terri Fellers <terri.fellers@ssbbartgroup.com>
> *Subject:* Re: Test harness action items
> 
>  
> 
> Hi Bryan,
> 
>  
> 
> Joanie will have a URL for the test results. I have one but I don’t 
> know if that will be the final one.
> 
>  
> 
> Best,
> 
> Rich
> 
>  
> 
> Rich Schwerdtfeger
> 
>  
>

Received on Friday, 28 April 2017 20:47:34 UTC